Automating Crop Canopy Data Collection for Crop Management: Difference between revisions

From Computer Laboratory Group Design Projects
Jump to navigationJump to search
No edit summary
Line 1: Line 1:
Contact: Michael Gifford, [[NIAB]] <Michael.Gifford@niab.com>
Contact: Michael Gifford, [[NIAB]] <Michael.Gifford@niab.com>


Models are used to optimise the production of potato crops by forecasting yield and scheduling irrigation. The models require regular data on the percentage of ground covered by leaves to quantify light interception and evapotranspiration.  Currently this is collected by users on the ground but this is time-consuming, expensive and often poorly performed. Automating the collection of data would improve the accuracy of model outputs and enable wider use by reducing the cost of operation.  Satellites with optical sensors can collect suitable data, but in Northern Europe, cloud cover prevents reliably collecting observations on the required weekly basis.
 
The EU’s Copernicus Programme provides free high resolution optical imagery and synthetic aperture radar (SAR) imagery from the Sentinel 1 and 2 satellites.  SAR imagery can be collected through clouds and during the night.  At present there is no available service for estimating canopy cover from SAR imagery.
Models to optimise potato crop production forecast yield and schedule irrigation using manually collected data on leaf canopy coverage to quantify light interception and evapotranspiration -- time-consuming, expensive and often inaccurate. Such data can be collected by satellite but optical sensing is impeded by cloud cover in Northern Europe. Synthetic aperture radar (SAR) imagery from the Copernicus Programme is collected through clouds and during the night but there is no available service for estimating canopy cover from SAR imagery. Your challenge is to develop a machine learning system   to estimate canopy cover from SAR imagery and integrate with existing  models.
The challenge for the teams will be to develop a system for estimating canopy cover from SAR imagery and to integrate this into the existing digital model for Potato Yield Forecasting. NIAB has extensive existing data sets of “ground truthed” canopy cover at defined locations which will be provided to the team.
We anticipate that the project will likely require expertise in machine learning, image analysis and manipulation techniques database development and the use of APIs.

Revision as of 21:49, 19 November 2019

Contact: Michael Gifford, NIAB <Michael.Gifford@niab.com>


Models to optimise potato crop production forecast yield and schedule irrigation using manually collected data on leaf canopy coverage to quantify light interception and evapotranspiration -- time-consuming, expensive and often inaccurate. Such data can be collected by satellite but optical sensing is impeded by cloud cover in Northern Europe. Synthetic aperture radar (SAR) imagery from the Copernicus Programme is collected through clouds and during the night but there is no available service for estimating canopy cover from SAR imagery. Your challenge is to develop a machine learning system to estimate canopy cover from SAR imagery and integrate with existing models.